Changelog

A small fix on top of 2.7.0.

Fixed

• Closing the app no longer looks like a crash. Quitting TraxxTool from a terminal with Ctrl+C, or letting the system ask it to close, could pop a "TraxxTool crashed" report even though nothing had actually gone wrong. Deliberate exits now close quietly and file nothing. Seen mainly on the Linux build.

The big one: Medley Studio, a whole new room in TraxxTool that joins songs from your library into one seamless medley and lands it in the Workbench as an ordinary project. Alongside it, Kemper players get everything the HeadRush players have had, and the click stops being one fixed noise you either take or leave.

Medley Studio (new)

A medley is now something you build in an afternoon instead of a weekend. Pick songs from your library, and TraxxTool works out an order, a tempo for each join, a key plan that keeps every song singable, and a crossfade on every track

• then hands you the result as one project with real stems, a real click, real

lyrics and real markers. It exports like any other song, so it reaches the iPad the same way everything else does.

• Play it, then edit it while it plays. The medley streams live, so tempo, key, levels, every crossfade handle and every lane mute change under your hands without stopping. There is no render-and-listen cycle.
• "Make a medley for me." Choose the songs, and you get a finished, playable medley to react to. Changing your mind about it is the easy part; starting from nothing was the hard part.
• Bars and beats, never milliseconds. Every join is anchored to a downbeat by default, measured off the real audio rather than a nominal tempo.
• A key plan that respects singers. Keys are chosen around the circle of fifths so consecutive songs are related, and nothing is pitched further than you allow. Chords and lyrics transpose with the audio.
• A tempo plan that respects arrangements. A ballad dropped into a run of up-tempo songs is put at half time rather than stretched to breaking, because 68 against 132 is a clean half-time pair, not a 94% stretch.
• The Doctor. A panel that reads the medley and says what will actually sound wrong - a segment stretched too hard, a key nobody can sing, a seam with no room to fade - and offers the fix as a button.
• A trim editor and a seam editor, with waveforms, so any join can be done by hand when the automatic one is not what you meant.
• Edit Medley, from the Workbench. A rendered medley is not a one-way door: the plan that made it is saved beside it and reopens.
• Touch-friendly: every scrollable pane in the tab takes a finger.

Medley Studio is brand new, and it is the largest single feature we have built. The first version of anything this size meets a far wider variety of real songs than any amount of testing covers, so you may hit rough edges. There is a banner in the tab with a Report a problem button attached to it - please use it. Your medley plan is saved separately from the songs it was built from, so nothing in this tab can damage your library.

Free and Pro. Building, playing and editing a medley is free at any length. Sending a finished medley of more than three songs to the Workbench needs Pro.

New

• Tempo and key for your whole library, looked up rather than guessed. Select songs in the Library and TraxxTool asks whoever actually knows: the backing-track site that built the song states both, the chord sites state the key, and only when nobody knows does it analyse the audio. Every value records where it came from, because "75 BPM because the people who made it say so" and "75 BPM because we measured it" are not the same claim. Free for up to 20 songs at a time; Pro sweeps the whole library in one run.
• Import a folder of audio as a voice. If your cues live in a folder - from a bandmate, an old library, or another tool - point TraxxTool at it and it becomes a proper voice sitting next to every other one. Filenames are matched to cues in stages, and anything it cannot place confidently is imported as a custom cue rather than dropped, with the whole thing shown as a review list before it commits.
• Guide Click. The Comp tool (the C button on a track) now has a third mode, for click tracks. Select the sections of the click you actually want in your ears, press GUIDE CLICK on the lane underneath, and those sections are summed into one extra exported track. The click track itself exports completely unchanged, so the drummer still gets the whole thing - the two just go to different buses. Several click tracks can each contribute and you still get one Guide Click out, because one is what a rig has a spare bus for. Its bus is set in the Librarian's Routing panel, next to the spoken cue bus.
• Click Sounds. The tick TraxxTool voices a click with is now a choice, not a fact. Eight built-in voices to start with - the stock Karaoke Version click, a woodblock, an electronic beep, drumsticks, a cowbell, a rim knock and two kit voices - and you can build your own from any audio clip you like. Settings > Workbench > Click Sounds...
• Kick and snare, and kick and hi-hat, for anyone whose click is a part rather than a metronome. A kit is far easier to lock to than a tick, because it sounds like the thing you are already following - which matters most through an intro, or anywhere the recorded drums drop out and the band has nothing to hold on to. Put one on the Guide Click bus and the drummer can still have the plain click on theirs.
• The click and the guide click choose separately. Keep the stock click on the drummer's bus and put something completely different in your ears.
• An editor for each sound, laid out like the cue recorder: a waveform, Level, Tone and Punch, and trim handles. Plus the one control that is specific to a click - the click point, which marks where inside the clip the beat actually is. Samples rarely start on their transient, and a sound whose attack is 12 ms in will drag the whole song until you set it. Press Detect and it usually gets it right on its own.
• Existing click tracks are left exactly as they came, by default. Turn on "Apply my click sound to existing click tracks" and TraxxTool finds every tick in the stem, works out which ones are accents, and re-voices them where they already are. Variable tempo is fine - the new ticks land on the old ones, to the millisecond. It runs in the background with the song still playable, and it never touches your files: the original stem stays exactly as downloaded and you can switch back at any time.
• Accents are worked out from the audio, not assumed. Whichever tick is rarer is the accent, whether it is louder, quieter or just a different sound, so a click sound TraxxTool has never seen still comes out with its downbeats in the right place. A click with no accents at all stays that way rather than having a downbeat invented for it.
• Regions, built from your lyrics. If your sheet has section headers, it already holds everything a region needs: the header is the name, its first line is where the region starts, and the next section is where it ends. Press Regions from Lyrics in the Tools menu and the whole set is built in one go - named, coloured by what kind of section it is, and in order. Every verse gets the same colour whether you called it "Verse", "Verse 2" or "Second Verse", because on stage the colour is there to say what a section is, not which one. Re-run it any time you change the sections; it replaces the set rather than half-merging it.
• A Regions editor, alongside Markers and Cut in the tool switch. Drag in the lane under the cue track to draw one, drag it to move it, drag its edges to resize. Double-click renames; right-click sets the colour and the playback mode. It is Stage Traxx's own thirteen colours in the picker, so the colour you choose here is the colour on stage, and regions cannot be made to overlap because Stage Traxx will not take them that way.
• Playback modes come across too: Continue, Repeat Once, Repeat, Skip and Pause. Set a chorus to Repeat and it loops until you move on; set a verse to Skip and the song plays without it.
• A glance, not a distraction. With Show All on, regions stay visible in every other tool as a thin coloured strip along the top - enough to read the shape of the arrangement while you work on something else. Show All off hides them.
• They export into the .st4b, so they are there when the song loads on stage.
• Kemper PROFILER support. A Kemper plugin sits beside the HeadRush one and does the same job: your Performances become the rig list, each song picks one, and its Program Change is sent when the song loads. Drop module toggles, MultiBlocks and Slot changes on the timeline and the amp follows the song hands-free. Built for the PROFILER Stage, and the Head, Rack and Player speak the same MIDI.
• The eight modules A B C D X MOD DLY REV are what "blocks" means on a Kemper, and the five Performance Slots plus the Stage's four Effect Buttons are what "footswitches" means.
• A MultiBlock lands exactly where it says it will. The Kemper is told the state you want rather than "flip this", so if you stomped a switch yourself mid-song the recall still comes out right.
• Delay and reverb toggles keep their tails by default, so a MultiBlock never cuts a reverb off mid-bar.
• Read your Performances off the board. TraxxTool asks the PROFILER for its Performance and Slot names, and for what sits in each module and whether it is on, so the timeline says "MOD - Rotary Speaker" instead of "Block 6". This needs the Kemper's MIDI OUT reaching the computer, which USB gives you for free; over 5-pin DIN it is a second cable. The panel says plainly which of those you have.
• A Kemper entry in MIDI Devices for anyone who would rather place plain device markers: every documented switch and pedal controller, named.
• Five more stage devices, ready to automate. Tick the ones you own in Rig > Devices & MIDI and each gets its own editor mode, with every control named so you never type a raw CC number:
• Line 6 Helix / HX (Floor, Rack, LT, HX Stomp and Effects) - snapshots, setlists and presets, the stomp footswitches, looper and tuner. A snapshot marker placed just after a preset marker lands both in one move.
• Neural DSP Quad Cortex - presets, scenes A to H, the eight footswitches, tap and tuner.
• Strymon TimeLine - presets, bypass, the A/B/TAP switches, the knobs and the looper.
• TC-Helicon VoiceLive 3 - every vocal and guitar block as an on/off, so a song can drop the harmony for a verse and bring it back for the chorus on its own.
• Behringer X32 / Midas M32 - faders, mutes, pan, mute groups and scene recalls, alongside the X Air support already here. Fader markers are written in dB and glide between values.
• And four more, beyond the guitar rack. Same deal: tick what you own.
• Allen & Heath Qu (Qu-16/24/32/Pac/SB) - mute any channel, DCA, mix or mute group from the timeline, and recall scenes. Killing the band except the click between songs is now one marker.
• Allen & Heath SQ (SQ-5/6/7) - all sixteen Soft Keys, so a song can fire whatever you have assigned to them, plus scene recall.
• Nord Stage 3 - the keyboard player's version of a rig change: programs, the organ, piano and synth sections, the drawbars, rotary speed and the effects.
• Elektron Digitakt - per-track level, mute, filter and sends across all eight tracks, plus pattern changes.
• Gear that listens on MIDI notes now works. Desks often take orders as notes rather than CCs, which TraxxTool previously could not send from a marker or write into the exported MIDI track. It can now, so the mutes above fire live and survive the export.
• QLab shows, read straight into TraxxTool. If you run QLab, the cues you have already set up to listen for MIDI can now fire from a song. Open the Rig tab's new QLab page, either read the workspace QLab has open (on the same Mac) or open a copy of the `.qlab4/.qlab5` file (on any machine), and press Use these cues. In the Workbench, pick that show in the editor-mode dropdown and right-click the timeline: your cues are right there, grouped by cue list, under their own numbers and names. One click places one. It sends exactly the message that cue is waiting for, live and in the exported MIDI track, so it still works when the set runs from Stage Traxx.
• Two shows, two songs, no clashes. Each workspace you import becomes its own device, so importing the second show doesn't replace the first. Both stay in the mode dropdown under their own names.
• Markers keep working whatever you import afterwards: a marker records the MIDI message when you place it, so it fires correctly even if that show's cues aren't loaded any more, or the song is opened on a machine that has never seen the show file.
• Every device's controls are in the Add Marker menu now. Picking a device as your editor mode and right-clicking the timeline used to offer one "<Device> control…" entry that opened a dialog. The controls are listed directly instead, grouped the way the device groups them, so a switch or a cue is one click. Faders still open the picker, because they need you to say what value.
• A warning when two devices would fight over the same message. Everything TraxxTool sends leaves by one MIDI output, and a device works out what is for it by MIDI channel alone - so two boxes on the same channel both act on everything, and you find out when the guitar rig changes patch halfway through a song. TraxxTool now says so instead: in the Workbench's MIDI panel when a song's own markers would clash, and in Rig > Devices & MIDI when two devices you have ticked are set to the same channel. Program Change gets special attention, because a device reacts to *any* Program Change on its channel whatever the number - so a rig change and a QLab cue on one channel collide even though the numbers differ.
• Controls… next to every device in Rig > Devices & MIDI. It lists everything that device understands and sends any one of them on the spot, so you can prove the cable, the port and the channel at soundcheck rather than mid-song. It works for your own custom device profiles too.
• Batch-edit a whole Stage Traxx backup. Next to the Merge controls in the Library there is now Batch Edit…: open any .st4b and change the songs already inside it, rather than only adding new ones. It opens instantly even on a huge library, because it reads the song data and leaves the audio alone.
• MIDI commands written into your lyrics, changed everywhere at once. If your set has patch changes embedded in the words - as Stage Traxx [midi:] tags rather than separate MIDI files - the editor lists every command it finds across the whole backup, with how many songs use each one. Pick one, say what it should become, and it changes in every song you selected. Move a channel, renumber a program, nudge every controller value, remove a command you no longer use, or add a load-time tag to every song at once.
• Song settings in bulk, with a Ripple mode that walks a value down the selection in the order shown - the fast way to give each song in a set its own program number instead of typing 40 of them.
• Lyrics find and replace that knows what it is looking at. It can tell a chord from a section header from a MIDI tag, so the default scope is the words alone and a three-letter search can't quietly rewrite your commands. Or scope it deliberately at chord symbols, section headers or tag lines. Chords written into the lyrics can be transposed, and MIDI tags, chords, ChordPro directives or repeated blank lines can be stripped out.
• Nothing happens without a preview. Every operation shows exactly which songs it would touch and, for lyrics, the changed lines before and after.
• A restore point before every batch, taken automatically. Roll back to any of them from the Restore point button. The file on disk is not touched until you press Save, and saving writes a new archive atomically rather than editing the original in place.
• Ending cues. Songs end as reliably as they start, and the band wants warning of the last section as much as it wants a count-in. New button Place Ending Cues, next to Place Intro Cues: park the playhead on the downbeat where the song actually finishes, press it, and you get the new End cue called a bar early followed by a 1-2-3-4 that lands the band on the last hit. One undo takes the lot back out.
• Two new cues to go with it: "End" and "Last Time". End is the end itself, as distinct from Outro, which announces the outro section starting and can be most of a minute earlier. Last Time is the last trip through a section. Both sit in Song Sections alongside Verse and Chorus, and in the recorder. In Spanish they are spoken "Fin" and "Última vez", against "Final" for Outro.
• The two-bar count-in. Under Count (1-8) in the Add Marker menu: "1 ... 2 ..." on beats one and three, then a full 1-2-3-4 on the next bar. It sits on your click's real beats and reuses the count cues you already have, so it costs nothing to record or generate.
• Count in eighths with Ctrl+Shift. Shift-clicking a count number has always laid the whole 1-2-3-4 across the click's beats; hold Ctrl as well and you get "1 & 2 & 3 & 4 &", each "and" landing halfway between the beats, with the trailing one after the last number. The menu labels change as you hold the keys so you can see what you are about to place. Uses one new cue, And ("y" in Spanish), which you are offered once.
• My cues. Custom cues used to be offered only if the song you had open already used them, which made a cue recorded for one song unreachable in every other. The Add Marker menu now has a My cues list holding every custom cue in your library, whichever voice or song it came from.
• Choose which cues to generate. The cue coverage grid in Voices and Cues is now tickable. Untick the ones you do not want and "Generate missing cues" covers only the rest, so wanting one cue no longer means paying for every gap. Everything starts ticked, so the old behaviour is one press away.
• Star the controls you actually use. An X32 profile carries over a hundred controls and a song uses six of them. Every control in a device's Rig page now has a star next to it. Starred controls sit at the top of the Add Marker menu for that device, above everything else, the first nine on Alt-1 to Alt-9. Stars are per device and stick around between songs and sessions.
• "In this song" for devices too. Under your favourites, the Add Marker menu lists the controls this song is already using, the same as the cue menu has always done for voice cues. Between the two lists, laying the seventh marker for a song is a keypress rather than a scroll.
• Favourite cues, the same way. Cues get stars too. Star one in the ★ column of Voices and Cues, or right-click a cue you have already placed and pick Favourite Cue, and it moves to the top of the Add Marker menu above Count and Song Sections, the first nine on Alt-1 to Alt-9. Stars belong to your library, so they are the same in every song and in every voice. A starred section cue still places as a section.
• Your own track types, properly managed. Adding a type like FX or Pads used to be a one-way trip: you typed a name and it was there forever. There is now a Track Types window that lists every type you have added and lets you rename, delete, and give each one an icon and colour from the app's own set. Custom types now read at a glance on the timeline instead of all sharing the same purple tag.
• It opens from wherever you happen to be looking at track types: the type menu on any track header, the Track Types… button in the Library's Bus Routing panel, a right-click on any row of that routing table (which opens straight onto the type you clicked), or Settings > Workbench.
• Deleting a type checks your library first. If any songs use it, it says which ones and how many tracks, and asks what those tracks should become - Other by default, or any other type including your own. It then rewrites those songs (keeping a backup of each) so nothing is left carrying a type that no longer exists. Previously those stems would have landed on the fallback bus at the next export with nothing to warn you and no way to correct them, because the type was gone from the routing table. Re-export the songs to send the new routing to Stage Traxx.
• Renaming carries your library with it, for the same reason: it says how many songs use the type, updates them all (backup kept), and takes the type's bus assignment with it. Say no and the rename is called off rather than leaving half your set on a name that no longer exists.
• Presets no longer move your own types. Switching bus routing preset used to quietly dump every custom type onto one shared bus, because presets only describe the built-in types. Your types now keep the bus you gave them, and each one gets its own line in Bus Details.
• The cue editors show you the cue you already have. Click any cue in Record Guide Cues and its saved audio is drawn straight away, with a Preview button that plays it. It used to show an empty pane whether or not the cue existed, so the only way to hear your own cue was to record over it first.
• Refresh Cue Audio, on the right-click menu of any cue marker. Record or generate a cue again, decide in the song that the new one is better, and the markers you already placed pick it up. Until now they kept the old audio and the only fix was deleting every marker and placing it again. Works on a whole selection at once, and runs by itself when you close a cue editor.
• Move where a cue lands on the beat. Peak-aligned cues fire on their loudest moment, which is right nearly every time - but not for a count spoken with a hard consonant, or a generated cue with a breath louder than the word. Both cue editors now draw that moment as an amber line you can drag, with the automatic one left behind as a ghost; double-click to go back to it. The point belongs to the cue's audio, so one drag fixes every song using that cue, and the export lands on exactly what you heard.
• An AI Cue Editor, the generated-voice counterpart of the cue recorder. Pick a voice, pick a cue, hear what it has, and change the words behind it: "Chorus!" is shouted where "Chorus." is called, and a couple of tries is normal. Regenerate replaces that cue's audio everywhere it is used, the wording is remembered per voice, and the same alignment handle is there. Open it from Settings > Guide Voice, or from a cue marker's right-click menu when you are on an AI voice.

Changed

• Moving a cue marker is instant. Dragging a cue used to leave its audio and waveform behind for seconds while every guide clip in the song was thrown away and re-decoded from disk - and the cues were silent for the whole rebuild, which is exactly when you wanted to hear them. Cue audio is now read once per file per session, a move slides the loaded clips instead of reloading them, and the rebuild that still happens keeps the waveforms already drawn. A busy song's cue rebuild went from over a second to under five milliseconds, and most edits no longer trigger one at all.
• Dragging a cue past the start of the song, which grows the count-in padding, no longer forces the slow path either. That was the one case that always hit it, and it is where cues get dragged most.
• TraxxTool no longer ships with any gear in it. A band with no HeadRush should not be carrying a HeadRush around. Every device is its own plugin now, downloaded from Settings > Plugins - fourteen of them, from HeadRush and Kemper to the Qu, the X32 and QLab - so you install the ones you actually own. Installing one puts it in Rig > Devices & MIDI to enable, gives it its own page in the Rig tab, and puts its controls in the Add Marker menu. Removing it takes all of that away again. Generic MIDI is always there.
• Nothing is taken away from you on update. If you were already using the HeadRush or X Air support, it is kept and works exactly as before - it just lives in your own plugins folder now, where you can update or remove it like any other. The Rig tab says so once.
• Older device markers quietly learn the MIDI message they resolve to the next time their song is opened, so even switching a device off later can't leave a marker sitting on the timeline firing nothing.
• The plugin catalog can be searched and filtered. With fourteen devices in it, Settings > Plugins > Browse now has a search box and a type filter, so you can look for "mixer" or narrow the list to Devices or App extensions. Every entry says which it is.
• Every device has its own page in the Rig tab, which is where you set it up. What is on the page depends on the device: a rig library and board sync for the HeadRush and Kemper, a show importer for QLab, and for gear that can't be read back, every control it understands with a Send button so you can prove the wiring at soundcheck.
• The Rig tab copes with a lot of devices. Pages wrap onto a second row, and past seven the rest collapse into a "More" menu instead of squeezing the panel off the screen.
• The Workbench's HeadRush inspector section is now called Rig, and the Clear menu's HeadRush entry is now Rig Markers, because both belong to whichever rig device your song uses. Nothing moved, only the label.
• The rig library now records which device each rig came from, so a HeadRush and a Kemper can live in it side by side without either sync disturbing the other's rigs.
• The Downloader now has separate "Purchased" and "Downloaded" columns. They answer different questions - what you have bought but not yet brought into your library, versus what you pulled down recently - so each sorts on its own and neither borrows the other's date. A blank Purchased cell means the source has not told us yet; if your library was fetched before TraxxTool started reading purchase dates, press Refresh Tracks and they fill in. The cell says so if you hover it.
• The click track's + button explains itself. Hovering it now says what it does: it copies the click's own first bar and puts it in front, which is how you give a download that starts cold a one-bar count-in. Press it twice for two bars.

Pro and Community

The rule we hold ourselves to: everything involved in preparing a song is free and stays free. Nothing that was free before this release is paid in it, and nothing ever will be. What Pro buys is working at *scale* - a whole library or a whole set at once - which is where the hours are, and which is the part worth paying for.

• Merging is free again, up to five songs at a time, as many times as you like. It used to be one free merge ever, which walled you off at the exact moment you had decided the workflow worked, and it needed an account so we could count it. Now nothing is counted: five or fewer per run is free whether you are signed in or not, whether it is your first merge or your fiftieth. Adding this week's new song to your library never costs anything. Moving a whole book in one go is the Pro version.
• If you already used your one free merge under the old rule, that no longer means anything. You have your small merges back.
• Building a medley is free at any length. Every part of the work - staging songs, ordering them, planning keys and tempos, editing every seam by hand, and playing the whole thing back - costs nothing, however long the medley is. Sending a finished medley of up to three songs to the Workbench is free too, as often as you like. Only sending a longer one needs Pro.
• Looking up tempo and key is free for up to twenty songs at a time. That is deliberately more than a set list, so preparing a whole gig costs nothing. Sweeping an entire library in one run is the Pro version.
• Saving a batch-edited backup is a Pro feature. Opening a backup, reading through it, running any batch operation and previewing exactly which songs and which lines it would change all stay free, so you can see the whole job done before deciding anything. Restore points are free too. Writing the edited archive out is the Pro part.
• Every device is free, permanently. All fourteen of the device plugins - HeadRush, Kemper, Helix, Quad Cortex, the Qu, the SQ, the X32, QLab and the rest - are free, and so is anything that keeps a show from going wrong: the MIDI channel conflict warnings, the Controls… test-send panel, favourites, every cue. We would rather be the thing your whole rig runs on.

Fixed

• One title bar, on every platform. Windows had started drawing its own title bar above TraxxTool's, so the app wore two - the second one grey, empty and taking a strip of screen for nothing. It is gone. Linux never had the custom chrome at all and now does: same header, same window buttons, and the window still moves, snaps, tiles and resizes exactly the way your desktop expects, because those are still handed to it rather than reimplemented. macOS is unchanged - the traffic lights stay where a Mac window keeps them.
• Importing a song no longer spends AI credits. Bringing stems in places the intro cues for you, and one of them is the song's own title. A title is different in every song, so unlike the count-in - which ships ready-made and is free - it has to be generated, which is how a brand new user could be told they had run out of credits on their very first import, having asked for nothing. Importing now prices that before it starts: if the title can't be generated it is simply left out, the count-in still lands, and nothing is spent. Pressing Place Intro Cues yourself still explains the shortfall and offers to fix it, because there you asked.
• Your own click track starts at the start. TraxxTool lines an import up by its ENDS, which is right for a downloaded set where the click is longer than the stems because it carries the count-in on the front. A click you built yourself usually has no count-in, so it is shorter - and aligning ends shoved it forward into the middle of the song by exactly the difference. Nothing is placed later than the stems now. Downloaded sets import exactly as they always did.
• Count-in markers can be nudged. With Snap on, dragging one barely moved it: a count-in cue sits on a click transient by definition, so any drag shorter than half a beat snapped it back onto the transient it came from and the marker looked stuck. Hold Alt while dragging to place a marker freely, snap or no snap.
• Count-in cues land in the right place on songs with a pitching tone. An "Intro Count" stem usually holds a note for the singer to come in on. That note has its own attack, which read as a click, and it sustains, which drowned the real ticks - so the automatic "1 2 3 4" ended up scattered on exactly the songs that needed it most. TraxxTool now separates the count from the tone and reads the ticks off the count alone. The stem itself is untouched, and if you do re-voice the click, the pitching note is put back underneath the new one rather than thrown away.
• The Comp lane says what it is. The lane under a comped track was labelled "SomeTrackName_bt", clipped mid-letter for any name longer than about nine characters. It now reads "Backing track" (or "Guide click"), fits the header, and the export filename - which really is <stem>_bt.mp3, and really does matter - is in the tooltip where an explanation belongs.
• Warping the click no longer chases its own tail. Snapping uses the click track's transients, so warping or time-shifting the click moved the grid it was snapping to and the tool never settled. The grid is now pinned to where the click was when the edit started, and goes live again when you let go.
• A comped backing lane stays in sync with a warped stem. In preview only, the _bt lane ignored its parent's warp and played at the original tempo underneath it. Export always had this right.
• Ampersands no longer eat the next letter. Anything with an "&" in its name - "Allen & Heath Qu", a song called "Rock & Roll", a folder called "Covers & Originals", a QLab cue called "Smoke & Lights" - showed on buttons, checkboxes and menus with the ampersand gone and the next letter underlined, because that is how Qt marks a keyboard shortcut. Names now say what they say. This covers cue names in In this song and My cues, device and control-group names in the Add Marker menu, your own track type names, the Rig tab buttons, and the vocal-track button in Lyrics Studio. It also fixes the count menu, whose eighths labels read "1-_2-_3" instead of "1-&-2-&-3" while Ctrl+Shift was held.
• Imported cue folders lost their custom cues. Importing a folder of audio as a voice reported success, and any cue that was not a standard name was then gone the next time you opened the app or changed song. The audio was written under a filename with two extensions that nothing could find. Cues now land where the rest of the app looks for them, and any that a previous version stranded are recovered automatically the next time you open a song.
• Imported WAV and FLAC cues are converted properly. They used to be copied under an .mp3 name without being converted, which played here but shipped to Stage Traxx mislabelled.
• Cues you never generated no longer count as missing. A cue name left behind by a cancelled or failed generation became a permanent row that every voice was short of, which inflated what "Generate missing cues" offered to charge you for. Only cues that actually have audio somewhere count now.

The follow-up to 2.6.0: the parts of that release that were only half-done. Spanish now reaches the whole app rather than a handful of screens, and the Voices and Cues manager does what it was always meant to.

New

• Spanish across the app, not just the main screens. Every screen, dialog, menu, button and message now goes through the translation layer - around 1,500 more pieces of text than 2.6.0 covered, including the Library, the Downloader, the recorder, Song Details, the rig editors and every settings page. Anything not yet translated still shows in English, so nothing is ever blank or broken.
• Cue markers are called cue markers everywhere. The rename now reaches the screens 2.6.0 missed, and so does the Settings toggle for anyone who prefers the old "guide" wording.
• The Voices table. The Voices and Cues manager now opens on a real list of your voices - how many cues each one holds and when you last used it - above the coverage grid.
• Fill several voices at once. Select any number of voices and one run generates every missing cue across all of them, with the total cost stated before anything is spent. Your own recorded voice is skipped with a note instead of blocking the run.

Fixed

• Custom loudness target stopped responding in a non-English interface. The dialog compared the translated label against English text; it now tracks the selection itself.
• The Cloud Backup button could stick on "Cancel backup" for the same reason.

For translators

tools/i18n_extract.py --missing es lists what still needs translating and --coverage es checks a catalog for placeholder mistakes. Catalogs live in src/core/translations/<lang>/, split by surface.

A big one, shaped almost entirely by user reports and requests. Guide markers are now called cue markers (switch back in Settings if you prefer the old word), voices become first class, lyric sync lands the ending, and the app starts speaking Spanish.

New

• Voices and Cues manager. Every voice you use - AI voices and your own recorded one - now keeps its own cue library, shared across all your songs. A cue generated or recorded once is reused everywhere, custom phrases included. Settings > Guide Voice > Manage Voices and Cues shows which voice has which cues and fills the gaps in one batch, with the cost stated before anything is spent. Switching voices no longer silently replays audio made by the previous voice.
• Import your own cue audio. Already have professionally recorded cue packs? The recorder's new Import Audio button takes WAV, MP3, M4A, FLAC or OGG straight into the same trim and save flow as a recorded take.
• The whole count, one click. Hold Shift on a number in the Count menu to lay the full count (1 through N) on consecutive click beats - each number lands on a real beat, not baked into one clip. The feature was always there; now the menu shows it to you.
• Spanish is arriving. Pick your language in Settings > Interface. The main screens, menus and messages come across first, with the rest following in upcoming releases. Spoken cues can already be generated in Spanish too (set independently of the interface language), and each language keeps its own cue audio per voice - so a bilingual show is a setting away.
• Custom track types. Add your own types like FX or Pads next to the built-in ones - they show up in the role menu, routing, AI track categorisation and exports.
• Auto key detection. The song details Key field can now fill itself from the song's chords, with a Detect button and a gentle suggestion when the field is empty. Your manual entry is never overwritten.
• Tidy with AI. Paste rough lyrics into the Lyrics Studio Text tab and let AI clean the layout and section order - in the song's own language. Nothing is applied until you review and hit Apply Text.
• Live Backup (Pro). Turn it on and every song you finish is added to a working StageTraxx backup as you go - the workbench button becomes Export Song, a banner shows how many songs are collected, and Finalise Backup writes the finished .st4b whenever you are ready. Made for big migrations.
• A separate bus for spoken cues, if you want it. Routing now offers "Send spoken cues to their own bus" for consoles that keep the click and spoken cues on different channels. Off by default; most rigs are happier combined.
• Colour tint. Pick the app's accent colour in Settings > Interface from a curated set, and your choice rides into StageTraxx backups so ST4 picks a matching tint.
• Downloads browser upgrades. A new Added column (when your provider library added the song, or when you downloaded it), Status sorting that puts fresh downloads first, Ctrl+C to copy the selected songs as a list, and a right-click menu that understands multi-selection.
• Custom voice IDs that just work. Type any ElevenLabs voice id and the app verifies it as you type, showing the voice's real name - so a typo can't cost you a broken cue later.
• Cifra Club chord sheets. A new optional chords plugin covers Spanish, Portuguese and English songs alongside the existing sources.

Fixed

• Lyric sync no longer drifts at the end of songs. The last stretch of a song is now anchored to the vocal itself and paced to the song's own tempo instead of a fixed guess - previously the final section almost always needed manual correction. Thanks Luis for the on-point report.
• Chords always match what you hear. A pitch-shifted song now applies the shift to its chords automatically on export - including when the shift is baked into the audio, where the chords are transposed in writing instead.
• Squished menus at small sizes. At low resolutions or high display scaling the top menus used to crush into an unclickable sliver. The header now measures what actually fits and folds cleanly into a menu button, at every window size and scale.
• Recorder honesty on every platform. No microphone found, an input that will not open, or an OS privacy setting feeding silence (Windows too, now) all show clear inline guidance instead of a vanishing toast - plus a "Hearing you" check so you know the mic works before you record.
• A crash closing the lyric sync results window right after using one of its copy buttons.

A Mac-focused round of fixes from real reports. Thanks to everyone sending diagnostics - it is exactly how these get found.

Fixed

• macOS: recording your own guide cues works properly now. Two separate problems, both fixed:
• The app never asked for microphone permission, so macOS quietly handed it silence and your take came out flat and empty with no error. TraxxTool now asks for the mic like it should, and if a take does come back dead-silent it tells you the likely cause and where to fix it (System Settings, Privacy and Security, Microphone) instead of pretending all is well.
• Mics and interfaces clocked at 48 kHz (AirPods, many USB interfaces) refused to record at all. TraxxTool now records at the device's own rate and converts the take for you.
• macOS: switching your output device no longer needs a restart. Connect AirPods or change the output in System Settings and playback follows, instead of staying glued to the old device.
• Honest guide cue generation. If spoken cues fail to generate (no voice credit, network down), the app now says which cues failed and why, instead of announcing success while your markers sit there silent.
• Sections survive a lyric re-sync. Re-running lyric sync used to orphan your section markers and stack a duplicate set on the timeline. Markers are now matched back up by name, anything stale is cleaned away, and the lyrics-to-timeline link keeps working.

Fixed

• macOS: window dragging now works even where 2.5.2 didn't manage it. Some Macs (notably on the newest macOS 26 updates) accepted our "move the window" request and then quietly ignored it, so the window stayed stuck. TraxxTool now moves the window itself whenever the system doesn't, so dragging the top bar works either way. If a window still won't budge, Help > Report a Problem now captures exactly what happened to your drag, so send one in and we can see it.

Fixed

• macOS: the window can be moved again. On some Macs the main window could not be dragged at all - it just sat where it opened. Dragging any empty part of the top bar now moves the window, and double-clicking it zooms, just like a normal Mac title bar.

Improved

• Honest lyric timing. When line times are estimates (typed or pasted lyrics that haven't been Auto-Synced or tap-synced), the Lyrics Studio now says so, and points at Auto-Sync - which listens to your actual lead vocal - or Tap Sync for lining things up properly. No more silently drifting karaoke lines that look like a bug.

New

• Song sections, linked to your timeline. The sections in your lyrics (Verse, Chorus, Bridge) and the section markers on your timeline are now two views of the same thing, kept in step for you.
• A guide marker can now double as a section. Right-click it and choose Mark as Section, or pick straight from the Song Sections part of the Add Marker menu. It still calls out its cue exactly as before, it just wears a section colour and now counts as a real song section.
• Type a header like [Chorus] in the lyrics editor and it lights up. When you are ready, one press of Add Section Markers puts a marker on the timeline for every section. Press it as often as you like, it only ever adds what is missing, never a duplicate.
• It works both ways: add a section on the timeline and the matching [Chorus] header appears in your lyrics; rename or delete one side and the other follows. Sections still export to Stage Traxx as headings above your words.
• Auto-Sync no longer drops section markers on the timeline behind your back. Sections arrive in the lyrics editor, and you place them on the timeline when you want. There is a switch for the whole behaviour under Settings > General, and a per-sync tick if you want the markers straight away.

New

• Import desk recordings - the Channel Mapper. Record your show or rehearsal straight on your digital desk (a 32-channel Behringer WING-LIVE file, an X32/M32 card recording, any multichannel WAV) and drop the one big file on TraxxTool. Instead of a DAW session just to bounce channels out, the new Channel Mapper turns it into a ready-to-mix song in one screen:
• Every channel gets a row: include it or not, pair it with its neighbour as one stereo stem, and name it. Names drive the same role detection as every other import - "Kick" lands on the drum bus, "Click" on the click bus.
• TraxxTool analyses the file first and suggests the mapping - silent desk channels come unticked, channels that move together arrive paired as stereo, and a sparse, spiky channel is flagged as the click - each with its confidence shown on the row. Nothing is decided silently; it's all yours to override.
• Gigabyte-class show files are streamed, never loaded whole - both the analysis pass and the split itself run as a single ffmpeg stream, and the split preserves your recording's bit depth.
• Layout templates. Your desk is patched the same every show, so save the mapping once ("My WING rig") and the next import is two clicks. Built-in starting points included: mono channels, stereo pairs, and a Jamzone-style band set (click + guide + the parts as stereo pairs).
• Free for everyone, like every import feature - Community builds unlimited songs from desk recordings.
• Behringer WING add-on. A new community add-on (Settings → Plugins) that connects the Channel Mapper to your WING console: one click fetches every scribble-strip name and colour over the network (read-only - it never changes the desk), so your imported channels are labelled exactly as they were at front of house. Includes find-my-WING discovery - no IP address needed on most networks. The built-in X Air integration gained the same power for XR12/16/18 desks.
• Desk plugins API. Plugin authors can now register a "desk names provider" - a small read-only adapter that teaches the Channel Mapper to fetch channel names from any console. The WING add-on is the reference implementation.

New

• Loudness normalization. Backing tracks downloaded from different places land at wildly different volumes; until now the fix was a trip through a DAW. Tick Normalize on the workbench's export bar and every export is levelled to a consistent loudness (-16 LUFS by default, the live-set convention; streaming and broadcast presets included), measured the way Logic and the streaming platforms measure it (ITU BS.1770 / EBU R128, validated against the standard's own test signals).
• TraxxTool measures the mix your audience hears: your saved faders, pans, EQ and mutes, minus the click bus and any stems routed to a bus you mute on the device (the "band plays these live" setup from Bus Routing). One gain is then applied to the whole song via the Stage Traxx song volume - your stems and mix balance are never touched, so un-muting the drums live when the drummer can't make it still sounds exactly like your saved mix, just levelled.
• True-peak protected: the gain never pushes the mix's true peak (the 4x-oversampled kind, catching inter-sample overs) past -1 dBTP. Transparent gain only, no limiter - a song without the headroom lands as close to target as it honestly can, and says so.
• Click Loudness… to see this song's measurement: integrated LUFS, true peak, the exact gain the next export will apply, and a per-stem list showing what was measured and why anything was excluded. Odd one out in the set? Give a song its own target, or switch normalization always-on or off for just that song. Your song fader still rides on top as an artistic offset.
• Stereo mixes get the same gain baked in, so rehearsal files sit level across the set too. Every export re-measures automatically - change the mix, and the next export is right without any extra step.
• Chords. The missing piece for chord players: TraxxTool can now put chords on your songs and export them to Stage Traxx, drawn above the words on stage exactly where they belong.
• Open the workbench's lyrics room and click Import Chords. Paste any chord sheet - a copy from a chords website or a ChordPro file both work. Every chord lands on the matching synced lyric line, so your chords inherit the timing your lyrics already have; nothing to line up by hand.
• Sheet section headers (Verse, Chorus, Solo) become real song sections, and named sections now export to Stage Traxx as proper headings - also true for sections you already had from lyric syncing.
• Instrumental passages come across too: a bar line like "| Em7 | G |" becomes a timed line you can drag on the sync map like any lyric, shown in the chord colour.
• Chords show above their lines on the sync map, the karaoke strip stays clean words, and the StageTraxx preview shows exactly what will export. Made a mess? Import Chords also has Remove All Chords, and everything is undoable.
• A chord strip on the timeline. Toggle Show Chords on the workbench toolbar and a slim bar appears above the ruler charting the whole song: one colour-coded block per chord, from each change to the next, scrolling and zooming with the timeline. The same chord always gets the same colour, neighbouring chords never share one, and labels step aside rather than pile up when changes are close together.
• Ultimate Guitar add-on. A new separately-installed plugin (Settings, then Plugins) turns the Lyrics room into Lyrics & Chords: Import Chords gains a search box that finds the best-rated chord sheet for your song on Ultimate Guitar and loads it for review with one click. Fetches happen only when you ask, and the add-on shows its own legal notice before install - like all community add-ons, it is separate from TraxxTool and used at your own discretion.

Fixed

• Lyrics now stay in time on cut or tempo-warped exports. If a song had a Song Cut or a baked tempo change, the exported lyric line times were only shifted, not remapped - so the words could drift against the audio (the MIDI cues beside them were already handled correctly). Lyric lines now ride exactly the same cut-and-warp mapping as the MIDI, lines inside a removed section are dropped instead of piling up at the seam, and the StageTraxx preview in the lyrics room shows the corrected times.

New

• TraxxTool now updates itself. No more trips back to the website every release: when a new version is out, the app can tell you what changed and update in place. You choose how in Settings, then General, then Updates: Install automatically (updates download in the background and install themselves when you quit), Notify me (a notification shows you what's new, and nothing downloads until you say so - this is the default), or Never check. Help, then Check for Updates always asks on the spot, whatever the setting. Every download is integrity-checked before it is ever run, and an update only ever applies across a restart - never while you are mid-song.
• The full changelog is now on the website. Every release's notes live at traxxtool.app/changelog - the same notes the update window shows you, with a deep link per version.

Fixed

• macOS: "TraxxTool is damaged and can't be opened" is gone. Some downloads reported the app as damaged and told you to move it to the Bin, and even right-click then Open would not get past it. The build process was compressing the app in a way that broke its signature; it no longer does. You get the normal one-time "unidentified developer" prompt instead, and right-click then Open works as documented.

Fixed

• Stage Traxx 4 could refuse to import an exported song ("Unexpected Error / typeMismatch"). If a song had its musical Key set, TraxxTool wrote that Key in a format Stage Traxx could not read, and it rejected the whole import. Fixed: the Key now exports in Stage Traxx's own format, as do the song Notes and the song-select MIDI trigger. Re-export the song on this version and it will import. (Already made a broken export on an earlier version? tools/repair_st4s.py repairs an existing file without re-exporting it.)

New

• A guided Interactive Tour. New in this release: open Help, then Interactive Tour, and TraxxTool walks you through itself with a spotlight over each part of the app as you go, importing stems, placing guides and markers, cutting a section, and running Prepare Song for real on a bundled demo song. Fresh installs (and anyone updating from before this release) get a one-time offer to take the tour on first launch. Start or restart it any time from the Help menu; if you step away partway through, it offers to pick up where you left off.

Changed

• "Export Song" is now "Prepare Song." The File-menu item, the button in the workbench and the Library's status column all read Prepare / Prepared now. It is the same action, rendering your stems and manifest into your library ready to package for Stage Traxx; the name just describes what it does before anything leaves the app.
• Faster launch. The branded splash now appears within a moment of opening the app instead of after a dead pause, and a chunk of startup work that used to run before the window showed has been moved off the launch path. Cold starts feel noticeably quicker.

Fixed

• macOS 26 (Tahoe): Detect Tempo and other busy actions no longer crash the app. On the newest macOS, putting up the "busy" wait cursor could abort TraxxTool with no error message, taking down Detect Tempo among other operations. The wait cursor is now handled safely there (macOS shows its own spinner instead), so those actions simply run.
• Linux: the app could fail to start when launched from a desktop icon. Started from a launcher, systemd or a container (anywhere without a controlling terminal), a startup check could throw before the window ever appeared. It now falls back cleanly and launches as expected.

Fixed

• macOS: launching no longer bounces the Dock icon for up to a minute. The app used to unpack itself on every launch; on first open (while macOS also verifies a new app) that could take a minute of bouncing with nothing in Activity Monitor, and clicking the icon again during the wait started a second copy - two TraxxTool icons. The app now ships fully unpacked inside the .app the way Mac apps normally do, so it starts in seconds, and opening it again just brings the running window forward.

Fixed

• macOS: repeated error windows then a crash. On recent macOS versions, opening almost any window (Settings, About, Lyrics Studio) could set off a cascade of error dialogs mentioning "WindowChrome" and eventually take the app down. A Windows-only piece of window styling was running on Mac and breaking half-initialised; it no longer runs there at all. Mac dialogs keep their standard title bars, exactly as before.
• The Karaoke Version and Genius add-ons could crash the app. The copies on the plugin server were outdated and failed against current TraxxTool, taking the Downloader tab down with them. The server now has fixed copies: open Settings, then Plugins, and update (or reinstall) the add-on. A broken plugin now shows a clear message instead of crashing the app.

New

• Installed plugins now keep themselves up to date. A few seconds after launch, TraxxTool checks the plugin database and quietly updates any add-on you have installed to its newest version (a small note appears when it does). Only plugins you chose to install are touched, downloads are integrity-checked, and you can turn it off in Settings, then Plugins.
• Help now has a User Manual entry. It opens the bundled PDF manual, or the current manual on traxxtool.app if your install does not carry one.

Fixed

• The stem-splitting engine download crashed on every install. Clicking Download in the engine dialog brought up a "Failed to execute script" error before anything was fetched, on all installed builds (the same fault also blocked the MIDI add-on installer). The installer's bundled copy of pip could not run inside the packaged app; it now ships in a form that can. If this hit you, just Retry the download on this version.
• Custom ElevenLabs voice IDs work without Advanced mode. Picking "Custom voice id" in the guide-voice list now reveals the ID field for everyone. It used to live with the bring-your-own-key controls, so credits users had nowhere to type it.
• Changing your Project Root now refreshes the Library immediately - no more manual refresh (or restart) to see the songs from the new folder.
• MIDI markers now show in the Lyrics Studio StageTraxx preview. With "Embed MIDI as lyric tags" turned on, the [midi] tag lines your timeline markers export are now part of the preview, exactly as StageTraxx will receive them. The Custom MIDI Commands window also lists them (read-only, they are edited on the timeline) so everything that exports is visible in one place.
• The library export list now says StageTraxx 4, so it cannot be mistaken for the new StageTraxx 3 option next to it.

New

• Export to Stage Traxx 3. Still on ST3, or share songs with a bandmate who is? Pick StageTraxx 3 in the export format dropdown. You get the same choices as ST4: a Song Share (.stsa) to import alongside your library, or a full Backup (.st3), with Merge to add songs to an existing backup without disturbing what's already there.
• Multitrack Backup collapses each output bus into one ST3 track (ST3 tops out at 8), bouncing the stems together and only pulling a track's level down if it would clip. If a song uses more than 8 buses it tells you to simplify the routing first, rather than quietly dropping stems.
• Single Mix Backup embeds one stereo Band Mix per song - click and cues in one ear, the backing in the other, the parts your band plays live left out - for a lightweight single-file backing track.
• Guide cues, markers, lyrics and MIDI all come across. Because ST3 has no attached-MIDI support, your device MIDI is written as timed lyric tags automatically.
• Backups and Merge are Pro, exactly like the ST4 versions; the song share is free.
• Embed MIDI as lyric tags instead of a .mid file (Settings → General). Off by default for ST4; always used for ST3. Handy if your setup reads MIDI from the lyrics rather than an attached file. The Lyrics Studio StageTraxx preview now shows your [midi] command tags inline.

Fixed

• First launch no longer hangs on the splash screen. On a brand-new install, the "Welcome to TraxxTool" window could open hidden behind the loading splash, which then never went away - the app looked frozen before you ever saw it. The welcome step now waits until the main window is actually on screen. If 2.0.1 stuck on the splash for you, this is the fix (and in the meantime, clicking the splash card gets you through).
• The macOS installer window shows its proper background again. The drag-to-Applications window could come up plain white with two bare icons; it now reliably shows the styled layout with the arrow and instructions, and the installer disk mounts with a TraxxTool drive icon.

Fixed

• KV plugin settings tab now appears after install even when Playwright hasn't been downloaded yet (deferred the browser-automation imports so register() runs unconditionally).
• ElevenLabs gating: a saved API key no longer grants access unless Advanced / BYOK mode is enabled. Previously a leftover key bypassed the credits gate.
• Rig tab device errors ("no module src.gui.device_connection") on the HeadRush and X Air panels: the module is now included in frozen builds.
• macOS deep link (traxxtool://) no longer opens a second instance; events that arrive during startup are queued and replayed.
• macOS custom chrome: the native title bar is now transparent with traffic-light buttons, matching the Windows custom chrome.
• macOS app icon now shows in Finder (was missing the .icns bundle).
• Windows taskbar icon fixed on machines that launched from an installer shortcut (AppUserModelID now set on the Inno Setup shortcuts).

Changed

• macOS builds ship as branded .dmg installers with an Applications shortcut and drag-to-install instructions instead of bare .zip archives.

The first public release of TraxxTool.

New

• Lyrics Studio. A room of its own for lyrics: type or paste them, see them exactly as StageTraxx will render them, and drag the sync map against the vocal stem until it lands. Exports drop .txt/.lrc sidecars next to the song.
• Stem separation, built in. Drop in an mp3 and get six stems back - free on your own machine, or on a cloud GPU when the laptop is busy. You can also pull a hard-panned click out of an old backing track.
• Community charts. Share and download finished arrangements. A chart now carries every parameter and is fingerprinted against the stems it was built from, so a download either matches your audio or tells you plainly that it does not. Rated, browsable in its own window, and never carries lyric text.
• Rig tab. Devices get their own room instead of being scattered through Settings: HeadRush, X-Air mixer, QLC lighting and generic MIDI gear, each with a connection banner so you always know whether you are editing live or offline.
• Tempo detection and normalising. TraxxTool works out a song's BPM and uses it everywhere - the click, the count-in, the DAW export. Normalising fits a drifting click onto a steady grid, and toggling it off puts the original back.
• Sign in without a Google account. Alongside "Continue with Google", you can now get a six-digit code emailed to any address - no password to set, remember or reset.
• Google Drive downloads, alongside direct links.
• Song naming that sorts itself out. Stem filenames are mined for the title and artist, with an AI guess as backup, and nothing is renamed until you have seen the review dialog.
• Cloud backup that actually restores. A complete content-addressed mirror of your library: re-uploading unchanged songs costs nothing, and a second machine can pull the whole thing back down.
• Multi-command markers. One marker can now fire several MIDI commands at once.
• Align Click picker. The one-click button guesses which stem is the click and which is the drums. When it guesses wrong - two click stems, or the Drums role on the wrong file - shift-click or right-click to say what you meant.
• Combined DAW export. Reaper and Ableton in one pass, carrying your MIDI through, with the Live set built from a real Live 12 template.
• Guide voices you can hear before you pick. A curated list with previews instead of raw model ids. Custom cues are global now, not per-song.
• Diagnostics and support. One place to see what the app thinks of your setup and send a support bundle. Telemetry is opt-in.
• Deep links. traxxtool:// links open in the copy you already have running instead of starting a second one that fights over the library.

Changed

• Credits by default. AI features run on credits with no provider account needed. Bring-your-own-key now sits behind a single Advanced switch rather than a setting per feature. Bulk jobs tell you the cost before they start.
• New look. Bundled fonts, a proper brand palette, app-drawn window chrome, compact/comfortable density modes, and a zoom-aware timeline scrollbar.
• Renamed to TraxxTool throughout. Existing installs migrate their settings, licence and library automatically on first launch.
• The website has been restyled and reshot, with per-platform downloads.

Fixed

• Smart quotes stopped being cleaned up. Track titles and artists carrying curly quotes were passed through untouched, which quietly hurt lyrics matching. The normalisation had been dead for some time and nothing failed loudly.
• Doomed AI jobs no longer start. Having AI configured is not the same as having the balance to pay for it; jobs used to begin and die at the API. The cost is now checked at the click.
• TraxxTool could hard-quit when you closed the workbench. Opening a song kicks off a background check for your HeadRush board; closing again within half a second of opening left that check running with nothing to hold it, and the app died outright a moment later - no error, no crash box, just gone. Most likely to bite when clicking through songs quickly, or with no board on the network.
• Self-record guide cues are never billed.
• Quitting during the first couple of seconds after startup no longer fires a model fetch at a half-closed window.
• Downloads survive a restart, and renaming one keeps folder and marker in step.
• AI settings are clearer. Bring-your-own-key controls (including the ElevenLabs key) all live on the AI tab now, credits is no longer offered as a "your own key" provider, and cloud stem splitting and ElevenLabs voice grey out cleanly when you are signed out instead of failing later.